Runtime monitoring for node.js applications
node-monitor records and gives you access to vital statistics about your running node.js application.
As a developer, node-monitor gives you a consistent pattern for exposing business events, monitoring exceptions, and managing errors and callbacks in your asynchronous code.
During configuration and deployment, node-monitor lets you configure standard output logging, or customized logging and alerts. Logging and alerts can be defined application-wide, module-wide, or for a specific event or error condition.
The provided loggers let you log to files, to a CouchDB database, to an email distribution list, an SMS gateway, an internal event, or an external program.
When your application exits, node-monitor produces an activity report listing activity for each monitor that fired throughout the lifetime of the application.
Node-monitor installs with npm and comes with an extensive suite of tests to make sure it performs well in your deployment environment.
To install and test node-monitor:
$ npm install monitor
$ npm test monitor
Start by defining the monitor at the top of your module. Subsequent examples use this monitor variable.
var monitor = require('monitor')('Customers');
Whenever you wish to record an event, call monitor.event():
monitor.event('Customer purchase', invoice.amount);
If you pass a Date object to the monitor, it assumes it's a start of an interval, and records the number of milliseconds from that time. Example:
var beforeDbSave = new Date();
db.save(customer, function() {
// Monitor the DB save time
monitor.event('Customer save time, ms.', beforeDbSave);
});
In the above example, customer database save times are monitored. Logging can be turned on or off at configuration time or at runtime. First, last, maximum, minimum, average, count, and total execution times are maintained for the Customer save time, ms. monitor.
Node-monitor lets you keep track of the health of your running node.js application by giving you a consistent pattern for trapping program errors and exceptions.
Errors are conditions preventing your function from producing the intended results. Here's an example using the try/catch pattern:
try {
... do something
} catch (e) {
monitor.error('Customer insertion failure', e);
...
}
In the above example, if an exception is thrown between the try and catch block, the Customer insertion failure monitor will log the error, and keep track of failure statistics.
Using the asynchronous callback pattern, you can monitor errors that occur in asynchronous functions, easily trapping and forwarding these errors to the callback function.
Long form example:
function saveCustomer(customer, callback) {
...
db.save(customer.id, function(err, dbObject) {
// Forward database errors to our callback
if (err) {
monitor.error('Customer db.save error', err);
if (callback) {
callback(err);
}
return;
}
...
});
}
Same, only using the short form:
function saveCustomer(customer, callback) {
...
db.save(customer.id, function(err, dbObject) {
// Forward database errors to our callback
if (err) return monitor.error('Customer db.save error', err, callback);
...
});
}
In these examples, the Customer db.save error monitor will log and track the error, and if a callback is specified, it is called with the error object as the first (and only) argument.
Monitor loggers are separated into two categories - eventLoggers and errorLoggers. These loggers can be configured application-wide, module-wide, or for individual event and error types.
The default event logger is require('sys').log, and the default error logger is require('sys').debug. Changing these defaults and writing your own loggers are done at configuration time (see Configuration below)
External alerts are written with loggers, and can be as simple or as complex as you desire. For example, you could write a logger which sends an email when the error occurs.
Email logger example:
// Email logger
var fs = require('fs');
var exec = require('child_process').exec;
var mail_cmd = "mail -s 'Program Error' john.doe@example.com";
var errorLogger = function(message, value, err, monitor) {
// Write the error message to a temp file
var tmpfile = "/tmp/" + Math.random() + ".out";
fs.writeFile(tmpfile, message, function (err) {
if (err) throw err;
// Mail the message
exec(mail_cmd + " < " + tmpfile, function(err) {
if (err) throw err;
// Remove the temporary file
fs.unlink(tmpfile, function(err){
if (err) throw err;
});
});
});
}
See the Configuration section below for attaching loggers to various monitors within your application.
node-monitor uses the config package for configuration. This allows you to specify configurations in a file or on the command line. See the node-config package for more information about specifying configuration options for your different application deployments.
The following configuration parameters are recognized:
* enabled - (boolean) Should the monitor be enabled? (default: true)
* eventLogger - Logger[s] to use for events (default: sys.log)
Can be a single logger, or an array of loggers.
* errorLogger - Logger[s] to use for errors (default: sys.debug)
Can be a single logger, or an array of loggers.
* maxLogSize - Limit the log output to this size (default: 10k)
These configuration parameters can be specified globally, or as defaults for all monitors in a particular module, or on a monitor by monitor basis within a module.
Global - specify them as defaults for the 'monitor' module. Example:
// production.js - Configurations for the production deployment
module.exports = {
monitor: {
defaults: {
eventLogger: null,
maxLogSize: 4196
}
},
...
}
The above configuration turns off logging for the event logger, and sets the maxLogSize parameter for all monitors (unless overridden).
Module Defaults - specify them as the 'defaults' monitor for your module. Example:
// production.js - Configurations for the production deployment
module.exports = {
Customer: {
monitor: {
defaults: {
errorLogger: require('monitor/emailLogger'),
maxLogSize: 4196
}
}
}
}
// Customer.js - Customer module
var config = require('config')('Customer');
var monitor = require('monitor')('Customer', config.monitor);
The above configuration defines module-level monitor defaults for Customer
Per Monitor - specify per-monitor configurations like per-module (above), only use the monitor name instead of 'default'. Example:
// production.js - Configurations for the production deployment
module.exports = {
Customer: {
monitor: {
default: {
errorLogger: null
},
'Customer insertion failure': {
errorLogger: require('monitor/emailLogger')
},
'Customer save time, ms.': {
enabled: false
}
}
}
}
// Customer.js - Customer module
var config = require('config')('Customer');
var monitor = require('monitor')('Customer', config.monitor);
The above configuration defaults error logging off for monitors within the Customer module. The the Customer insertion failure monitor overrides this default, and the Customer save time, ms. monitor is disabled.
When you run require('monitor')('my-module', config), an instance of a ModuleMonitor class is returned. Module monitors contain individual monitors for your module, and methods for easily adding to the monitors.
node-monitor
require('monitor')(moduleName, configuration) - This returns a Module monitor object for the specified moduleName, creating it if necessary.
moduleName - The name of your module configuration - An object containing these elements: enabled - (boolean) Should the monitor be enabled? (default: true) eventLogger - Logger[s] to use for events (default: util.log) Can be a single logger, or an array of loggers. errorLogger - Logger[s] to use for errors (default: util.debug) Can be a single logger, or an array of loggers. maxLogSize - Limit individual log output to this size (default: 10k)
require('monitor').getAllMonitors() - This returns an object containing named ModuleMonitor objects for each module.
ModuleMonitor
Instances of this class contain a dictionary of Monitor objects for each named monitor in your module. Convenience methods exist for easily adding to individual named monitors.
new ModuleMonitor(moduleName, configs) - Constructor
Inputs: moduleName - Name of your module configs - A dictionary of monitor configurations for this module key: monitorName, or 'defaults' for the module level defaults value: A configuration object for the monitor, possibly including: enabled - (boolean) Should the monitor be enabled? (default: true) eventLogger - Logger[s] to use for events (default: util.log) Can be a single logger, or an array of loggers. errorLogger - Logger[s] to use for errors (default: util.debug) Can be a single logger, or an array of loggers. maxLogSize - Limit individual log output to this size (default: 10k) Output: A new ModuleMonitor class instance
ModuleMonitor.get(monitorName) - Get a named Monitor object. If the Monitor object exists it will be returned, otherwise it will be created.
Inputs: monitorName - Name of the monitor Output: An instance of the Monitor class for the specified monitorName
ModuleMonitor.getMonitors() - Get the list of Monitor objects for the module.
Inputs: (none) Output: An object containing all monitor objects for the module, by name.
ModuleMonitor.event(name, value, data) - Monitor an event. This adds an amount to the specified monitor, and logs the event.
Inputs: name - The event (monitor) name. value - A numeric value to add to the monitor. Default = 1. If this is a Date object, the number of milliseconds between the Date object and Date.now() is added to the monitor. data - An optional object to pass on to the event logger. May be specified as the second argument in place of value (in which case value will default to 1).
Output: The named monitor object (for chaining)
ModuleMonitor.error(name, value) - Monitor an error that shouldn't be occurring. This monitors and logs the specified error. It can be used for exception processing as well as asynchronous error processing.
Input name - The error monitor name error - An object representing the error callback - An optional method to call (passing the error) after logging.
Output: monitor - This monitor (for chaining)
Monitor
new Monitor(name, moduleName, config, moduleConfig) - Constructor.
Inputs: name: This monitor name moduleName: Name of the containing module config: Specific configurations for this monitor: enabled - (boolean) Should the monitor be enabled? (default: true) eventLogger - Logger[s] to use for events (default: console.log) Can be a single logger, or an array of loggers. errorLogger - Logger[s] to use for errors (default: console.debug) Can be a single logger, or an array of loggers. maxLogSize - Limit the log output to this size (default: 10k) moduleConfig: Default monitor configurations for the module enabled - (boolean) Should the monitor be enabled? (default: true) eventLogger - Logger[s] to use for events (default: console.log) Can be a single logger, or an array of loggers. errorLogger - Logger[s] to use for errors (default: console.debug) Can be a single logger, or an array of loggers. maxLogSize - Limit the log output to this size (default: 10k) Output: A new Monitor object
Monitor.getHits() - Return the number of hits this monitor has recorded. Hits are the total number of logEvent() and logError() calls.
Monitor.getTotal() - Return the total values this monitor has accumulated. Values are specified on logEvent() calls. logError() calls accumulate a value of one (1) for each call.
Monitor.getAvg() - Return the overall average for the monitor. The average is the total amount as reported by getTotal() divided by the number of hits as reported by getHits().
Monitor.getMin() - Return the smallest value added via the logEvent() or logError() methods.
Monitor.getMax() - Return the larges value added via the logEvent() or logError() methods.
Monitor.getFirst() - Return the first value added via the logEvent() or logError() methods.
Monitor.getLoggers() - Return an object containing loggers added using addLogger(). The keys are the logger IDs, and values are the logger functions.
Monitor.getLast() - Return the last value added via the logEvent() or logError() methods.
Monitor.getConfig() - Return the actual configuration used for this monitor. This is a mixin of the program defaults, module defaults, and monitor configs passed in to the constructor.
Monitor.getName() - Return the name of this monitor.
Monitor.getModuleName() - Return the module name this monitor was created under.
Monitor.isEnabled() - Returns true if the monitor is enabled, false if disabled.
Monitor.enable(enabled) - This enables or disables the monitor. Disabling prevents the monitor from accumulating values and logging messages.
Input: enabled - (boolean) Enable or disable the monitor
Monitor.reset() - Reset the monitor accumulators to their original state.
Monitor.addLogger(loggerFunction) - Attach a logger function at runtime.
Loggers added using addLogger() will be called for all error and event logging called on this monitor.
Input: loggerFunction(message, value, data, monitor) - A function to run when an event or error is logged. The function accepts: message - A formatted message for logging value - The numeric value of the event data - The data object associated with the event monitor - A reference to this monitor object for accessing monitor data
Output: loggerId - An ID associated with this logger so it can be retrieved using getLogger(), and removed using removeLogger().
Monitor.removeLogger(loggerId) - Remove a logger that was added using addLogger(). This removes a logger by the ID assigned using addLogger().
Input: loggerId - The ID returned by the addLogger function when adding the logger.
Monitor.logEvent(value, data) - Log and accumulate an event.
Input:
value - An optional numeric value to add to the monitor. Default = 1.
If this is a Date object, the number of milliseconds between the
Date object and Date.now() is added to the monitor.
data - An optional object to pass on to the event logger
Output:
monitor - This monitor (for chaining)
Monitor.logError() - Monitor an error that shouldn't be occurring.
Input error - An object representing the error callback - An optional method to call (passing the err) after logging. Output: monitor - This monitor (for chaining)
Released under the Apache License 2.0
See LICENSE file.
Copyright (c) 2010 Loren West